• The Korean Journal of Applied Statistics
• /
• v.35 no.2
• /
• pp.177-188
• /
• 2022

The support vector machine (SVM) has been successfully applied to various classification areas with a high level of classification accuracy. However, it is infeasible to use the SVM in analyzing massive data because of its significant computational problems. When analyzing imbalanced data with different class sizes, furthermore, the classification accuracy of SVM in minority class may drop significantly because its classifier could be biased toward the majority class. To overcome such a problem, we propose the DOC-SVM method, which uses divide-oversampling and conquers techniques. The proposed DOC-SVM divides the majority class into a few subsets and applies an oversampling technique to the minority class in order to produce the balanced subsets. And then the DOC-SVM obtains the final classifier by aggregating all SVM classifiers obtained from the balanced subsets. Simulation studies are presented to demonstrate the satisfactory performance of the proposed method.

• Proceedings of the IEEK Conference
• /
• 2000.09a
• /
• pp.119-122
• /
• 2000

인터넷이 일상생활에 다양하게 활용되면서 인터넷 채널을 통한 정보의 형태는 문자와 이미지 외에 음성, 오디오 신호 및 동영상 부분까지 확대되고 있다. 본 논문에서는 MPEG4-CELP를 인터넷 화상 통신의 음성 코덱용으로 사용하기 위한 최적화 기법 및 알고리듬의 개선을, DSP칩이 내장된 보드가 아닌 인터넷의 터미널로 사용되고 있는 펜티엄 프로세서를 장착한 PC에 초점을 맞추어 수행하였다. MPEG4-CELP VM C소스를 분석 및 프로파일(Profile)한 결과를 토대로 패라미터 추출을 위해 많은 연산을 수행하는 부호화기에 대해서 CPU상에 부하를 많이 주는 함수들을 제 1차 최적화 대상 함수들로 선정하고, CPU에 부하를 많이 주지는 않으나 호출되는 회수가 많은 함수를 2차 최적화 대상 함수로 선정해, C소스 레벨의 소프트웨어 파이프 라이닝(Software Pipelinging) 기법들을 적용하여 최적화를 수행하였다. 또한 1차 최적화 대상 함수의 경우에는 소프트웨어 파이프라이닝의 적용과 함께 연산량 감소를 위한 알고리듬 변형까지 수행하였다. 위의 과정을 거쳐 최적화 된 MPEG4-CELP는 펜티엄Ⅲ 450㎒ PC에서 음성을 부호화 하는데 원 VM소스에 비해 약 2배정도의 시간이 단축되는 것을 확인하였다.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.39A no.3
• /
• pp.140-148
• /
• 2014

In this paper, We proposed Gaussian function as a basis of hybrid method. Hybrid method is to extrapolate late time and high frequency data using early time and low frequency data. This method takes advantages of both MOT and MOM as well as having shorter running time and smaller error. For this method a better basis function is required. We compared the performance of the result with proposed function and conventional basis including Hermite and Laguerre polynomial.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2004.11a
• /
• pp.473-476
• /
• 2004

본 논문에서는 프라이버시를 침해 하지 않는 데이터 마이닝에 대해 다룬다. 방대한 데이터에서 유용한 정보를 추출하는 데이터 마이닝분야에서 데이터로부터 프라이버시 보존의 중요성이 부각되고 있다. 그래서 프라이버시의 침해를 막기 위한 방법으로 실제 데이터를 사용하지 않고 잡음이 들어간 데이터를 사용한다. 그리고 프라이버시를 침해하지 않기 위해 잡음이 들어간 데이터로부터 데이터의 확률 밀도 함수(PDF)만을 복원한다. 이렇게 복원된 확률 밀도 함수만을 이용하여 데이터 마이닝기술, 예를 들면 분류화에 곧바로 적용함으로써 프라이버시를 보존하는 것이다. 하지만 분류화에 사용되는 데이터의 1차원적인 확률 밀도 함수만 가지고는 군집화에 사용하기가 부적절하다. 따라서 본 논문에서는 군집화를 하기 위해 잡음이 들어간 데이터로부터 결합 확률 밀도 함수(Joint PDF)를 복원하고, 복원된 결합 확률 밀도 함수만 가지고 군집화를 할 수 있는 방법을 다룬다.

• Journal of the Korea Society of Computer and Information
• /
• v.3 no.3
• /
• pp.63-71
• /
• 1998

This paper is to suggest designing and implementing tool by transforming source code to meta language. We use formula, which is based on relationship between variables and functions, in class extraction and restructuring method in other to extract most appropriate class.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2001.04a
• /
• pp.421-424
• /
• 2001

본 연구는 디지털 오디오데이터에 워터마크를 삽입하고 추출하는 알고리즘을 제시한다. 삽입방법은 3단계의 웨이브릿 변환을 행한 후 가장 낮은 주파수대역으로부터 2번째 주파수 대역에 워터마크를 삽입하는데 그 워터마크는 키에 의한 의사난수를 이용한다. 추출시는 상관함수(correlation function)을 이용하여 추출한다. 잡음에 대한 필터 특성 또한 보인다.

• Proceedings of the Optical Society of Korea Conference
• /
• 2003.07a
• /
• pp.296-297
• /
• 2003

본 연구에서는 푸리에 변환법에 의한 위상정보 추출 기술을 개발하고, 주파수 영역에서의 창함수 필터에 따른 위상추출 특성을 분석하였다. 푸리에 변환법은 위상이동법과는 달리 정현파 패턴이 투영된 하나의 영상만을 이용하여 3차원 형상정보를 추출할 수 있는 장점이 있다. 획득된 영상은 오일러 공식으로부터 다음과 같이 표현할 수 있다. (중략)

• Journal of Korea Water Resources Association
• /
• v.51 no.3
• /
• pp.235-246
• /
• 2018

The main purpose of this study is to suggest a methodology for identifying vulnerable region in Choyang creek basin susceptible to soil losses based on runoff aggregation structure and energy expenditure pattern of natural river basin within the framework of power law distribution. To this end geomorphologic factors of every point in the basin of interest are extracted by using GIS, which define tractive force and stream power as well as drainage area, and then their complementary cumulative distributions are graphically analyzed through fitting them to power law distribution to identify the sensitive points within the basin susceptible to soil losses with respect to scaling regimes of tractive force and stream power. It is observed that the range of vulnerable region by scaling regime of tractive force is much narrower than by scaling regime of stream power. This result seems to be due to the tractive force is a kind of scale dependent factor which does not follow power law distribution and does not adequately reflect energy expenditure pattern of river basins. Therefore, stream power is preferred to be a more reasonable factor for the evaluation of soil losses. The methodology proposed in this study can be validated by visualizing the path of soil losses, which is generated from hill-slope process characterized by local slope, to the valley through fluvial process characterized by drainage area as well as local slope.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.15 no.5
• /
• pp.599-604
• /
• 2005

This paper presents fuzzy rules to predict diagnosis of Wisconsin breast cancer with minimized number of feature in put using the neural network with weighted fuzzy membership functions (NEWFM) and the non-overlap area distribution measurement method. NEWFM is capable of self-adapting weighted membership functions from the given the Wisconsin breast cancer clinical training data. n set of small, medium, and large weighted triangular membership functions in a hyperbox are used for representing n set of featured input. The membership functions are randomly distributed and weighted initially, and then their positions and weights are adjusted during learning. After learning, prediction rules are extracted directly from n set of enhanced bounded sums of n set of small, medium, and large weighted fuzzy membership functions. Then, the non-overlap area distribution measurement method is applied to select important features by deleting less important features. Two sets of prediction rules extracted from NEWFM using the selected 4 input features out of 9 features outperform to the current published results in number of set of rules, number of input features, and accuracy with 99.71%.

• Journal of KIISE
• /
• v.44 no.10
• /
• pp.1019-1025
• /
• 2017

Binary similarity analysis is used in vulnerability analysis, malicious code analysis, and plagiarism detection. Proving that a function is equal to a well-known safe functions of different versions through similarity analysis can help to improve the efficiency of the binary code analysis of malicious behavior as well as the efficiency of vulnerability analysis. However, few studies have been carried out on similarity analysis of the same function of different versions. In this paper, we analyze the similarity of function units through various methods based on extractable function information from binary code, and find a way to analyze efficiently with less time. In particular, we perform a comparative analysis of the different versions of the OpenSSL library to determine the way in which similar functions are detected even when the versions differ.